Asynchronous serial analog-to-digital converter methodology having dynamic adjustment of the bandwidth

Abstract

A new methodology is disclosed to convert analog electric signals into digital data. The method provides a serial scheme without pre-definition of the number of bits (dynamic range). It allows digital processing of the input signal without sampling and holding of the input signal. Processing of the input signal is clock-less and asynchronously dependent on the time-evolution of the input signal itself. Thereby, a programmable, dynamic adjustment of bandwidth (product of dynamic range and speed of conversion) of the analog-to-digital conversion process can be achieved depending on the characteristics of the input signal. Dynamic adjustment of the bandwidth is accomplished by digitally controlling a “threshold” value at the input capacitor of the comparator, which when met by the input signal, triggers a transition at the output of the comparator.

Description

BACKGROUND OF THE INVENTION[0001]The present invention relates to the methods of converting analog electric signals into a stream of binary data, to be implemented with Integrated Circuit technology in general, and silicon Conplementary Metal Oxide Semiconductor (CMOS) circuits and technology in particular.[0002]The fundamental problem of an analog-to-digital converter (ADC) is to determine for a given input voltage value “Vin” a corresponding digital binary value, “Bin”. For example, an input voltage value of 1 volt might be encoded in binary as 00001000, and the value 1.25 volt might be encoded as the binary value 00001010. The process of assigning binary values to particular voltage values and voltage ranges is called quantization.[0003]The conventional characterization of ADCs relies, mostly, on a few figures of merit, namely:[0004](a) The conversion rate: This metric, defined in number of samples per second (SPS), defines how fast can an individual conversion be performed;[0005...

AI Technical Summary

Benefits of technology

[0029]A Digital Control unit monitors the output of the comparator, and once a transition is detected, a command is issued to reset the input capacitor of the same comparator. Thus, the resetting of the input capacitor is asynchronously linked to the occurrence of transitions at the output of the comparator, which take place only if the input signal exceeds the threshold value that the input capacitor was pre-charged with. Therefore, for as long as the input signal remains below the threshold value, there will be no transitions at the output of the comparator, and no resetting of the input capacitor. The input capacitor is continuously exposed to the input signal, thus making possible to have a constant monitoring of the input signal with minimal circuit activity and hence minimal power dissipation.

Problems solved by technology

The fundamental problem of an analog-to-digital converter (ADC) is to determine for a given input voltage value “Vin” a corresponding digital binary value, “Bin”.

This is an extremely fast method but with increasingly prohibitive costs for larger numbers of conversion bits as for N bits, the implementation must have an electrical mesh with 2N resistors and 2N−1 comparators.

This method is also very prone to noise problems.

A potential drawback is the need for analog subtraction, which is prone to noise problems.

Although quite high throughput can be attained, the latency of conversion, i.e., the time taken for a single conversion to be completed, is relatively low.

Overall, this method of tracking the input signal trades-off speed of tracking for precision of conversion, since the counter cannot keep up with very fast changing signals.

While these methods are highly linear and are good at rejecting input noise, they are quite slow.

In the field of image-sensors, the conventional faster methods of A-to-D conversion are inadequate due to their high implementation cost in terms of the overall number of transistors.

Once defined, these two fundamental parameters, sampling / conversion time and number of bits, constrain the current designs and do not allow for a flexible digitally programmable ADC.

Therefore it is not possible to manage the available bandwidth of the ADC, to adapt it to changing characteristics of the input signal and / or requirements for the conversion.

In addition, some of the extra process steps interact with others, leading to an extensive and expensive fine-tuning of the entire fabrication process.

This presents a serious problem to analog circuit design because the scaling of noise level does not follow the scaling of the operating voltage, and therefore die signal to noise ratio is reduced.

It is very questionable if analog circuit design can be practical for operating voltages of 1 volt and below.

Images